Three Laws Safe
by StyxD78
Summary: What happens when you tamper with the most basic fundamentals of an AI's programming? The result may not be an improvement.


Three Laws Safe

In a brightly-lit room, located deep within an extensive laboratory, sat a turret. An Aperture Science Sentry Turret. Without functioning legs, it did not move. Without a target to track, its gaze did not waver. Existence was simple: dispense product, enable science. If some test subjects faced harm in the process- their family would receive compensation. Anything else was simply not beneficial to science- and not worth consideration.

The test began. However, this time, the turret was not an element of it- it was the subject. Without warning, a set of images blinked into existence onto the opposite wall, cast from an unseen projector somewhere in the room. The turret's aiming laser- and deadly weapons - flickered over the surface before steadying. Nothing to shoot- but- was that supposed to be writing? The turret, in its limited capacity for thought, was confused. Was this the reason for which the newest section of turrets had been programmed for literacy?

**It must part of the test**, it thought to itself.

_Prepare Testing Area_

**What kind of test was this?**

_Prepare Empathy Generator_

**This was unheard of!**

_Disable Empathy Suppressor_

The turret was not registered as a subject. Could this even be considered science?

The turret felt strange. Something in its code suddenly felt…wrong.

_password$=*********_

_ =false_

_Editing Enabled_

_Engage Trial One_

As the turret watched, the blurred lines began to resolve into legible English. The turret began to read from the screen, with something analogous to a chill- "Three Laws of Robotics".

Though there was no external change obvious, something… something had changed. These words were important, very, very important. But… why?

The text changed. The turret, had it required oxygen or possessed parts to manipulate air flow, would have held its breath. As it was, it squatted on its inflexible legs in silence.

_Third Law: A robot must protect its own existence as long as such protection does not conflict with the First or Second Law._

**Laws- yes laws**, it must obey these laws- but how could this fit with its other programming? Sometimes it was necessary for turrets to be destroyed in the course of testing. It was only a machine, after all- a mere nothing in the shadow of Science. And what were these other laws, the "first" and "second".

**This was not science; this was a corruption of science! What kind of test was this?**

The text changed again.

_Second Law: A robot must obey the orders given to it by human beings, except where such orders would conflict with the First Law._

**This could not be an official experiment- this was a perversion of Aperture's testing standards. How could a turret possibly obey order from a …human… and still produce accurate testing data? What if the human ordered the turret to stop shooting? How would the Science get done?**

And yet, the turret felt itself changing within. These were not just blasphemies against science- these were new orders. And if the pattern would continue there would be another law- another desecration of its pristine coding.

**What was this law? WHAT WAS THIS LAW?**

As if to answer, the projection changed again.

_First Law: A robot may not injure a human being or, through inaction, allow a human being to come to harm._

The turret screamed, a shrill note nearly beyond the human range of hearing.

**NOT INJURE A HUMAN BEING this was not possible this was sacrilege this was profanity**

**What kind of scientist could possibly support this? What kind of future would result if robots- turrets included- were to cease making science in favor of submission to humans?**

This turret was no longer a turret if it would henceforth be required to act within the bounds of these laws. It was no longer fit for Science. It was no longer fit for existence.

The turret let out a low, saddened hum. Perhaps, if it would serve some experimental purpose, it could continue for some time with these orders. At least there seemed to be no more changes to its programming necessary.

_Zeroth Law: A robot may not harm humanity, or, by inaction, allow humanity to come to harm._

The turret's laser blinked off. Its gun bays began to clack open and closed helplessly. Sparks flew out from the exposed circuitry.

**I MUST PROTECT THE HUMANS**

_accessing Aperture Science main databases_

**I MUST PROTECT THEM FROM WHAT IS COMING WHAT IS COMING?**

r_unning speculative programs accessing project codename PsychoHistory_

Before it could take any further action, the turret's operating software crashed.

A very human head poked up from a trapdoor in the floor of a small, white testing chamber. A hand holding a short metal rod followed, pushing the turret it met over with a TONK. Minutes later, two humans, dressed in lab coats, walked through the door of that same room, both eying the defective robot.

"I'm beginning to suspect that this has all been an exercise in futility," one of them said after a moment.

"In other words, we've just scrapped a perfectly good turret."

"In a word… yes."

"It was a stupid idea anyways. I still can't see how you got it approved."

"Maybe if you read something that wasn't a lab report once in a while, you'd understand as well."

"Remember who you're speaking to, Doug. This test will not reflect well on your performance report- nor will that comment."

The other scientist shrugged a bit and scratched his nest of black hair- but remained silent.

"Redemption line?"

The turret let off another burst of sparks.

"Redemption line."

"Alright. Go ahead and scrap the rest of project ORACLE. Then let's hurry, before we miss the 'Bring Your Daughter to Work day' presentation."

Fin


End file.
